When I compile qt334 with the actual snapshot of gcc41 (20050813) I get an ICE
with -funsafe-math-optimizations. This ICE is new, the snapshot from last
week is working.
Michael Cieslinski
g++41k -c -funsafe-math-optimizations -o gllandscape.o gllandscape.ii
opengl/gllandscape.cpp: In member function 'void
()':
opengl/gllandscape.cpp:457: internal compiler error: tree check: expected
real_cst, have integer_cst in const_binop, at fold-const.c:1512
Please submit a full bug report, with preprocessed source if appropriate.
g++41k -v
Using built-in specs.
Target: x86_64-unknown-linux-gnu
Configured with: /gcc-4.1-20050813/configure /usr/local/gcc41k --
program-suffix=41k ,c++
checking
Thread model: posix
gcc version 4.1.0 20050813 (experimental)

Additional Comments From micis at gmx dot de 2005-08-17 07:09
Created an attachment (id=9510)
()
preprocessed source

Additional Comments From belyshev at depni dot sinp dot msu dot ru 2005-08-17 10:21
I could not reproduce this one, can you try current cvs HEAD ?

Additional Comments From micis at gmx dot de 2005-08-17 10:38
Sorry I can't. I sit behind a firewall with only http/https enabled, ssh is
blocked.
Michael

Additional Comments From reichelt at gcc dot gnu dot org 2005-08-17 11:23
This is most probably a duplicate of PR23391.
So I'm closing it now.

You might want to test the patch there to see if it fixes your problem.
If it is not fixed, please reopen the bug.

This bug has been marked as a duplicate of 23391

Additional Comments From micis at gmx dot de 2005-08-17 15:59
I reopen the bug because there is an other source file with the same ICE which
is not fixed by #c5

Additional Comments From pinskia at gcc dot gnu dot org 2005-08-17 16:06
Reducing.

Additional Comments From pinskia at gcc dot gnu dot org 2005-08-17 16:25
Confirmed reduced testcase:
double transport_sumexp(int numexp)
{
int k,j;
double xk1 = 1.0;
for(k=1; k<=numexp;k++)
for(j=1;j<=3;j++)
xk1 += 1.0;
return xk1;
}

Additional Comments From sebastian dot pop at cri dot ensmp dot fr 2005-08-17 23:23
Subject: Re: [4.1 Regression] ICE: tree check: expected real_cst, have integer_cst in const_binop, at fold-const.c:1512

I'm testing this patch on amd64 and i686. I will commit it once
validated.

Index: tree-chrec.c

RCS file: /cvs/gcc/gcc/gcc/tree-chrec.c,v
retrieving revision 2.24
diff -d -u -p -r2.24 tree-chrec.c
tree-chrec.c15 Aug 2005 12:26:07 -00002.24
tree-chrec.c17 Aug 2005 23:01:09 -0000
@@ -539,6 +539,9 @@ chrec_apply (unsigned var,
if (dump_file && (dump_flags & TDF_DETAILS))
fprintf (dump_file, "(chrec_apply \n");

+ if (TREE_CDE (x) == INTEGER_CST && SCALAR_FLAT_TYPE_P (type))
+ x = build_real_from_int_cst (type, x);
+
if (evolution_function_is_affine_p (chrec))
{
/* "{a, +, b} (x)" -"a + b*x". */

Additional Comments From cvs-commit at gcc dot gnu dot org 2005-08-21 10:59
Subject: Bug 23433

CVSRT:/cvs/gcc
Module name:gcc
Changes by:spop (AT) gcc (DOT) gnu.org2005-08-21 10:59:15

Modified files:
gcc : ChangeLog tree-chrec.c
Added files:
gcc/testsuite/gcc.dg/tree-ssa: pr23433.c

Log message:
PR tree-optimization/23433
* tree-chrec.c (chrec_apply): Translate INTEGER_CST to a
REAL_CST when the type is SCALAR_FLAT_TYPE_P.

Patches:

Additional Comments From pinskia at gcc dot gnu dot org 2005-08-21 14:53
Fixed.